i wrote this a couple weeks back, but i forgot to send it.

that's a good thing, too, because there have been recent
developments, which i have only now just caught up on...

but first, read this...

***

i see it's time for another quarterly report on fadedpage.

fadedpage.net is roger frank's experimental creation of
a collab digitization site, a la distributed proofreaders...
it's been up for 9 months now.

there's a lot to like about this site -- a lot to like -- so
i wish i could report that it's going strong.  but it's not.

it did grow steadily for the first 6-7 months, but lately,
it seems to be shrinking back.  it typically has around
a half-dozen people active in the last-24-hour-period,
and on average about a dozen active in the last week...

on the positive side, most of the kinks have now been
ironed out, and the workflow seems to be fairly steady.
the site has been brave about trying out new stuff, and
a number of things  have proven themselves as valuable.

among these are a "roundless" approach, with a certain
flavor of its own, as developed by roger, so that's good.
he quickly abandoned -- in practice, if not in his head --
some of the questionable aspects he had had originally,
and then settled into a system that was quite acceptable.

another innovation was "match the scan" methodology
that's a refreshing shift from the tired pseudo-markup
kludged together over time at distributed proofreaders.

roger has also been good about coding features that are
requested by his users.  one of those is _global_search_,
which is something that i think d.p. _still_ doesn't have,
despite many people having suggested it for years now.

roger leveraged the roundless mode such that _diffs_
were available quite quickly to users, which makes 'em
entirely more useful than they typically are over at d.p.

roger also trusted his users, deeply.  one example of it
is that any user could _immediately_ put a word on the
"good word" or "bad word" lists used by the spellchecker.
over at d.p., a user can only "suggest" a word, which then
must be approved by a higher-up, which is a bit insulting,
not to mention inefficient.  this implicit trust in the user
goes a long way, and it speaks volumes to the community.

so, all in all, fadedpage can be judged as a big success...

but it's hard for me to get too excited about fadedpage,
mostly because it has failed to attract a critical mass yet.
it's a big success.  but it's not a big community -- at all...

i'm not sure if that's because roger doesn't want one yet,
or if d.p. has sucked all the energy out of this sphere, but
whatever the cause, it's frustrating for the name of change.

***

it's also ironic that roger seems to constantly "discover"
points that i've been making repeatedly over the years...

for instance, here's something current from his forums:
>   I could have an application run locally
>   on each proofer's machine, connecting to fadedpage
>   through XML/RPC for pages. The local application
>   would be better than anything I could do in a browser.
>   The user's app would check out a page
>   and all processing would be locally. When done,
>   the page would be returned to fadedpage.
>   This is conceptually exciting and I've done some
>   preliminary testing of the concept with a DP user.
>   There are some technical hurdles but it's still
>   a potential project for down the road.

yeah, i've been saying that for years...  i've even coded
the app.  but i'm glad that it's "conceptually exciting"...

it's also frustrating that roger seems not to have absorbed
some of the lessons that he _had_ learned along the way...

here's another something recent in the forums:
>   scanno: Van Home for Van Horne fairly consistently.
>   Marked "Home" as a bad word, "Horne" as a good word.

those of you who have been here for years will recognize that
i would just make that a global change done in preprocessing.
this little bad-words/good-words dance, forcing volunteers to
find and fix each individual instance of the error, is ridiculous.

roger is also wasting a lot of his time trying to create a "master"
format that can churn out the various kinds of output he wants.
you're reinventing the wheel, roger.  i've already done that work.
a master format already exists; it's called zen markup language.

oh yeah, there's one more thing that bugs me about fadedpage,
which is that roger's answer, for may things the volunteers ask, is
"don't worry about that, the post-processor will take care of it"...

so, on the one hand, he's built a system that empowers volunteers
-- by giving them the power to take a page from o.c.r. to "final" --
but then, on the other hand, rips the new empowerment from them
by constantly reminding them that _someone_else_ will be making
the _real_ decisions _after_ their work is done.  it's a sad irony, it is.

roger will not achieve the breakthrough that he's looking for _until_
he decides to do the bulk of the "automatic" work in _preprocessing_,
and then lets his volunteers seize the power to move the work to final,
thereby _eliminating_ the need for any "postprocessing" at all...  voila!

***

in closing, though, i repeat that fadedpage has been a _success_...

however, it has failed to attract critical mass.  this is a problem...

i don't know what to do about this problem.  in the past, i have
suggested that project gutenberg funnel volunteers to fadedpage,
and i know that michael was willing to do that.  but i don't know
what roger thought about the idea.  and i don't know why it never
seemed to materialize.  fadedpage has done a good job of testing
some new ideas, and these experiments have paid off, but it won't
do anybody any good if nobody ever learns from these experiments.

let's hope someone catches on quickly...

-bowerbird
_______________________________________________
gutvol-d mailing list
gutvol-d@lists.pglaf.org
http://lists.pglaf.org/mailman/listinfo/gutvol-d